6. Conclusions
The foregoing examples demonstrate that the SRI provides
a valuable counterpart to the SPI for drought monitoring
and management. Whereas the SPI describes the climate
anomalies in isolation from their hydrologic context,
the SRI directly describes the effects of climate anomalies
on current hydrologic conditions as governed by land surface
physical processes, as least to the extent that they are represented
in modern hydrologic models. One strength of the
SRI is that for short accumulation periods (e.g., 1-month
and 3-month), it has utility where the SPI must be used
with caution. Another is that predictions of SRI can be
made that depend not only on climate, for which seasonal
prediction skill is generally low, but on hydrologic initial
conditions, which in some seasons are highly skillful predictors
for runoff (e.g., spring snow state in the western US). A
third strength is that calibrated runoff simulations are more
widely available for real-time application than naturalized
streamflow observation, which precludes the use of stream-
flow in a real-time framework (e.g., Modarres, 2007) in most
SHUKLA AND WOOD: VALUE OF SRI X - 5
locations. A weakness of the SRI is that it is based on modeled
runoff that cannot be verified everywhere (which is also
true of most current drought indices), and it reflects the
customary uncertainties associated with models. Model calibration
results can help by providing an indication for many
locations of the quality of the runoff simulation, hence the
reliability of the SRI.
We suggest that the multi-period SRI framework may
be more recognizable than percentiles to the drought research,
monitoring and management communities that have
adopted the SPI for a wide variety of applications. Although
from a technical standpoint, the difference between the SRI
and runoff percentiles is one of nuance rather than content,
the SRI may help bridge the gap between the science and
management communities by expressing content that is familiar
to scientists in terms that are familiar to managers
and decisionmakers.
Acknowledgments. The authors are grateful for the helpful
suggestions of Steve Burges; to Doug Lecomte, Victor Murphy
and Mark Svoboda for their feedback on the potential for hydrologic
modeling in drought monitoring; and for the funding support
from the NOAA Transition of Research Applications to Climate
Services program under Cooperative Agreement NAXXXXXXX.